Alex Alex - 2 months ago 10
R Question

Perfectly align several plots

My aim is a compounded plot which combines a scatterplot and 2 plots for density estimates. The problem I'm facing is that the density plots do not align correctly with the scatter plot due to the missing axes labeling of the density plots and the legend of the scatter plot. It could be adjusted by playing arround with

plot.margin
. However, this would not be a preferable solution since I would have to adjust it over and over again if changes to the plots are made. Is there a way to position all plots in a way so that the actual plotting panels align perfectly?

enter image description here

I tried to keep the code as minimal as possible but in order to reproduce the problem it is still quite a lot.

library(ggplot2)
library(gridExtra)

df <- data.frame(y = c(rnorm(50, 1, 1), rnorm(50, -1, 1)),
x = c(rnorm(50, 1, 1), rnorm(50, -1, 1)),
group = factor(c(rep(0, 50), rep(1,50))))


empty <- ggplot() +
geom_point(aes(1,1), colour="white") +
theme(
plot.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.border = element_blank(),
panel.background = element_blank(),
axis.title.x = element_blank(),
axis.title.y = element_blank(),
axis.text.x = element_blank(),
axis.text.y = element_blank(),
axis.ticks = element_blank()
)


scatter <- ggplot(df, aes(x = x, y = y, color = group)) +
geom_point() +
theme(legend.position = "bottom")

top_plot <- ggplot(df, aes(x = y)) +
geom_density(alpha=.5, mapping = aes(fill = group)) +
theme(legend.position = "none") +
theme(axis.title.y = element_blank(),
axis.title.x = element_blank(),
axis.text.y=element_blank(),
axis.text.x=element_blank(),
axis.ticks=element_blank() )

right_plot <- ggplot(df, aes(x = x)) +
geom_density(alpha=.5, mapping = aes(fill = group)) +
coord_flip() + theme(legend.position = "none") +
theme(axis.title.y = element_blank(),
axis.title.x = element_blank(),
axis.text.y = element_blank(),
axis.text.x=element_blank(),
axis.ticks=element_blank())

grid.arrange(top_plot, empty, scatter, right_plot, ncol=2, nrow=2, widths=c(4, 1), heights=c(1, 4))

Answer

Using the answer from Align ggplot2 plots vertically to align the plot by adding to the gtable (most likely over complicating this!!)

library(ggplot2)
library(gtable)
library(grid)

Your data and plots

set.seed(1)
df <- data.frame(y     = c(rnorm(50, 1, 1), rnorm(50, -1, 1)), 
                 x     = c(rnorm(50, 1, 1), rnorm(50, -1, 1)), 
                 group = factor(c(rep(0, 50), rep(1,50))))

scatter <-  ggplot(df, aes(x = x, y = y, color = group)) + 
                geom_point() +  theme(legend.position = "bottom")

top_plot <- ggplot(df, aes(x = y)) + 
                geom_density(alpha=.5, mapping = aes(fill = group))+
               theme(legend.position = "none") 

right_plot <- ggplot(df, aes(x = x)) + 
                geom_density(alpha=.5, mapping = aes(fill = group)) + 
                coord_flip() + theme(legend.position = "none") 

Use the idea from Bapistes answer

g <- ggplotGrob(scatter)

g <- gtable_add_cols(g, unit(0.2,"npc"))    
g <- gtable_add_grob(g, ggplotGrob(right_plot)$grobs[[4]], t = 2, l=ncol(g), b=3, r=ncol(g))

g <- gtable_add_rows(g, unit(0.2,"npc"), 0)
g <- gtable_add_grob(g, ggplotGrob(top_plot)$grobs[[4]], t = 1, l=4, b=1, r=4)

grid.newpage()
grid.draw(g)

Which produces

enter image description here

I used ggplotGrob(right_plot)$grobs[[4]] to select the panel grob manually, but of course you could automate this

There are also other alternatives: Scatterplot with marginal histograms in ggplot2